www.advmat.de www.MaterialsViews.com
Xingcheng Xiao,* Peng Lu, and Dongjoon Ahn There have been intense efforts devoted to promote vehicle electrification to create a diverse energy supply, reduce the dependence on fossil fuels, and address environmental issues. One of the critical issues in these efforts is the availability of sustainable energy storage systems, ideally lithium ion batteries with increased energy and power densities, for hybrid electric vehicles (HEV), plug-in hybrid electric vehicles (PHEV), and electric vehicles (EV). Despite much progress, the requirement of over 10 years lifetime for lithium ion batteries still remains a big challenge. The aging mechanism has been extensively investigated during the past decade, which is generally believed to be due to mechanical degradation and/or chemical degradation.[1,2] The mechanical degradation is mainly associated with the fracture of electrode materials especially for those with high energy density, eventually leading to the loss of electrical contact. Many efforts have been devoted to develop nanostructured electrode materials to avoid the mechanical degradation and improve the cycling stability of lithium ion batteries.[3,4] However, the high surface area of nanostructured materials normally leads to the formation of a large amount of solid electrolyte interphase (SEI), which results in a substantial first cycle irreversible capacity loss by trapping lithium in the SEI layer.[3] On the other hand, the chemical degradation is mainly due to the side reaction between the electrode surface and electrolyte, and the instability of solid electrolyte interphase.[5] The SEI is indispensable as a passivation layer in lithium ion batteries as it protects the electrode from co-intercalation of the electrolyte solvent molecules and prevents further decomposition of electrolyte.[6] Major efforts on improving the electrode service life are mainly focused on the stabilization of the SEI in order to reduce capacity loss in the first cycle and concurrently provide protection to the electrodes. Surface coatings have been shown to be an effective approach to mitigate both the mechanical and chemical degradation, therefore improving the reversible capacity, first cycle coulomb
Dr. X. Xiao General Motors Global Research and Development Center 30500 Mound Road, Warren, MI 48090, USA E-mail:
[email protected] Dr. P. Lu Trison Business Solutions Inc. 17 Bank St. Leroy, NY, USA Dr. D. Ahn University of Kentucky 179 F. Paul Anderson Tower Lexington, KY, USA
DOI: 10.1002/adma.201101915
Adv. Mater. 2011, 23, 3911–3915
efficiency, cycling behavior, rate capability and overcharge tolerance.[7,8] Various processes, such as sol-gel, precipitation, and chemical vapor deposition, have been developed to create artificial passivation layers on the surface of active material particles. The ideal coating would be ultrathin in order to avoid introducing too much ohmic resistance, at the same time, be conformal and continuous to protect electrode being attached from the electrolyte. However, it is difficult to precisely control the coating thickness and uniformity with those processes. Atomic layer deposition (ALD) is a unique process that can achieve conformal coating with several atomic layers and excellent controllability. Recently, it has been reported that direct deposition of oxide coatings on composite electrodes using the ALD technique could improve the cycling stability more efficiently because it provides a protective interface between electrode and electrolyte without sacrificing the electronic conductivity between the active materials and the conductive binder.[9–11] Our recent results also clearly demonstrated that a few nanometer thick oxide coating could significantly improve the cycling stability of lithium titanate composite electrodes even at a cutoff voltage as low as 1 mV.[12] Several papers have been published that provided an understanding of how surface coatings could affect the cycling performance of conventional electrodes.[7] However, very little work has been done to understand their role on the formation of the SEI layer and the transport process of lithium ions through the coatings. Part of the reason is the complexity involved in characterizing the real electrodes. In this study we have designed a simple thin film electrode system which does not involve the complicated porous structure, the polymeric binder and carbon black as used in the conventional composite electrode. Also, the flat electrode surface makes the surface composition analysis much simpler and more precise, since composition only evolves in one dimension through the thickness direction. Specifically, we selected amorphous Si as the active material and controlled the film thickness to be far below the critical size for fracture of around 300 nm.[13] Since the constraint from the substrate will limit the in-plane volume expansion, this will prevent the Al2O3 coating from breaking during cycling. We also inserted and removed a similar amount of lithium ions in the active materials through the surface, both with and without the Al2O3 coating, in order to compare the formation of the SEI layer on the electrode surface. Figure 1 compares the cycling behavior of Si and Al2O3/Si films as the negative electrodes under CC-CV conditions, where Si is around 75 nm thick and Al2O3 is around 5 nm thick. We purposely stopped the experiment in order to investigate the SEI formation after it becomes stable within 11 cycles. Although the Al2O3/Si sample has lower specific capacity than Si, it has
© 2011 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
wileyonlinelibrary.com
COMMUNICATION
Ultrathin Multifunctional Oxide Coatings for Lithium Ion Batteries
3911
www.advmat.de
mechanism will be discussed later with the support from XPS results. 4000 Si 3rd Figures 1c and 1d compare the dQ/dV in 1st Al2O3/Si 2 1st 2nd 2nd 90 the first three cycles between these two samCE-Si 3500 3rd CE-Al2O3/Si ples. There are three peaks in the discharge 3000 80 curves for pure Si: One around 0.6 volt cor1 responding to SEI formation, the other two at 2500 1st 2nd 3rd low potentials corresponding to lithium inser70 1st 2nd 3rd 0 tion into the silicon.[14] After the first cycle, 2000 0 1000 2000 3000 4000 2 4 6 8 10 12 the intensity of the peak corresponding to -1 Cycle number Capacity /mAh.g the electrolyte decompositions decrease, indicating that a stable SEI has already formed 8000 (d) 8000 (c) on the top surface. For the Al2O3/Si, the peak Si Al2O3/Si around 0.6 volt was not observed as shown 4000 4000 in Figure 1d, indicating that the formation 0 of the SEI layer was probably suppressed by 0 –4000 the Al2O3 coating. The coating is believed to 1st 1st act as a protective layer to prevent electrolyte –4000 –8000 2nd 2nd 3rd decomposition which consumes the lithium. 3rd –12000 However, we also observe an increase in –8000 0 1 2 3 0 1 2 3 resistance for the sample with Al2O3 coating. Voltage / V vs. Li/Li+ Voltage /(V vs. Li/Li+) As shown in the Figure 1b, the voltage hysteresis between charge and discharge is larger for Al2O3/Si than that without Al2 O3 coating. This is consistent with the dQ/dV analysis where the 1st reduction peak shifts to a lower voltage (Figure 1d) as compared to the Si without the coating (Figure 1c). The increase in oxidation-reduction peak separations shown in Figure 1d is a typical representation of the effect of kinetics due to the increase in resistance. On the other hand, both kinetics and storage capabilities are much better than Figure 1. A comparison of cycling behaviors between Si and Al2O3/Si electrodes. (a) Cycling the first 2 cycles. All these results suggest stability and Coulombic efficiency; (b) Voltage profile of the first three cycles under CC-CV that the coating becomes more ionically concondition; (c) dQ/dV curves of Si; and (d) dQ/dV curves of Al2O3/Si. (e) SEM images of cycled ductive during the first few cycles. Si and (f) Al2O3 coated Si. Figures 1e and 1f compared the surface morphology of Si and Al2O3/Si thin film electrodes after 11 cycles. Many microcracks are shown in the a slightly higher Coulombic Efficiency (CE) as indicated in uncoated Si thin film, probably due to the large volume change Figure 1a. For the first discharge cycle shown in Figure 1b, during the cycling. In contrast, the Al2O3/Si sample showed the capacity of pure Al2O3/Si is about 900 mAh/g when the only a few cracks, although both samples went through similar potential reaches the cutoff voltage, which is much lower cycling tests with very similar amounts of lithium inserted into than the value of 3800 mAh/g for Si. The continued insertion the film. A comparison of these surface topographies indicates of Li under constant voltage increases the capacity to about that the Al2O3 coating prevents the mechanical degradation of 3000 mAh/g for Al2O3/Si, but not much increase to Si. The lower the Si, which also explains why the Al2O3/Si sample has better capacity of the former can be attributed to the insulating oxide cycling stability and higher Coulombic efficiency than Si alone. layer, which causes the voltage drop and, therefore, increased The SEM images, combined with XPS and SIMS, prove that the overpotential to drive lithium into Si. A similar phenomena the Al2O3 coating remains on the Si thin films despite the has been observed in Al2O3 coated graphite and LiCoO2.[6] Given huge volume expansion, and the coating stabilizes the interface enough time, the capacity from the Al2O3/Si sample should between the electrode and electrolyte. The improved Coulombic have a similar value as the uncoated one. Interestingly, the efficiency can be partially attributed to the less crack generated capacity under constant current kept increasing during the 2nd during the cycling, where the fractured surface will form a new and 3rd cycle and then became stable, suggesting that the Al2O3 SEI layer. coating becomes more and more ion conductive in the first few SIMS was used to characterize the SEI thicknesses and comcycles. Further characterization indicates that some structural positions to compare the SEI formed on the Si and Al2O3/Si changes have occurred to the Al2O3, which facilitates lithium electrodes. Figures 2a and 2b show the chemical maps of the ion diffusion. This is a clear indication that the Al2O3 coating is various species observed at different depths in the SEI layer. going through a formation process where an insulating Al2O3 Briefly, Li2F+ indicates LiF; Li2O+ may be generated from both material can be activated to become ionically conductive. The 3
4500
3912
wileyonlinelibrary.com
(b)
Si Al2O3/Si
Voltage / V
100
dQ/dV / mAhV–1
dQ/dV / mAhV–1
Specific Capacity / mAh.g–1
(a)
CE / %
COMMUNICATION
www.MaterialsViews.com
© 2011 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
Adv. Mater. 2011, 23, 3911–3915
www.advmat.de www.MaterialsViews.com
and salt reduction:[18,19] 3Li + +PF− 6 + 2e = 3LiF ↓ +PF3
(2)
LiF was detected in the SEI layer on both Si and Al2O3/Si electrodes. Salt decomposition seems to be a more possible mechanism since Al2O3 could prevent electrons from reaching the electrolyte to reduce LiPF6. It is well known that LiPF6 is unstable and easily decomposes with heat and moisture.[20] On the other hand, Li2CO3 has been considered from the decomposition of electrolyte.[14,15]
COMMUNICATION
(1)
LiPF6 LiF(s) + PF5 (g)
(3) Since the insulating Al2O3 will prevent the electrons from reaching the surface to decompose the electrolyte, this may be the reason why much less Li2CO3 was detected on the Al2O3/Si surface. Since less electrolyte decomposition will mitigate gas generation in the cell, this will be very helpful for improving the performance of the cell. Less SEI formed on Al2O3/Si surface is beneficial for improving the Coulombic efficiency. XPS experiments were conducted to analyze the chemical binding state of Al at different depth. Figure 3a shows the binding energy of Al2p at different depth. At the top surface, no Al2p is detected because of the coverage of SEI layer. After ∼1 nm, two Al2p peaks appear, one at 75.5 eV corresponding to Figure 2. SIMS composition analysis of SEI layers. (a): chemical mapping on Si; (b): chemical Al2O3 and the other at 77.6 eV to AlF3.[21] The mapping on Al2O3/Si; (c): comparison of SEI composition between Si and Al2O3/Si close to deconvolution of the peak (Figure 3b) leads to the interface. The scale bar showing in (a) and (b) is 10 μm. an additional weak Al2p peak around 73.4 eV assigned to LiAlO2.[22] The relative peak area ratio (or the atomic ratio) of these three peaks at different depth Li2O and Li2CO3 which may decompose to Li2O upon sputis also shown in Figure 3c. Most of the LiAlO2 compound is tering;[15] P− is from P containing species in the SEI; Si+ and found at the top surface, which becomes weaker after 1 nm. Al+ are from the Si substrate and Al2O3 coating layer, respecLiAlO2 has been shown to possess excellent lithium ion contively. For the Si electrode, the SEI thickness was around 20 to ductivity (up to 3 × 10−5 Ω−1 cm−1) due to the partially occupied 30 nm, as indicated by the Si+ signal. The relatively high inten+ + Li ion sites inside.[23] The presence of LiAlO2 at the top surface sities of Li2F and Li2O signals indicate that this SEI consists might help reduce the energy barriers for Li+ ion insertion and mainly of LiF, Li2O and LiCO3. In contrast, the thickness of enhance the charge transfer kinetics.[24] After being sputtered the SEI layer on Al2O3/Si electrode was only a few nanometers with 1 nm from the top surface, the ratio between AlF3 and as indicated by the Al+ signal. Also, the most detected species Al2O3 remains almost constant. The AlF3 coating layer could were LiF and P containing species. Figure 2c shows the surreduce the formation of LiF films which may increase the elecface SIMS spectra of the SEI on Si and Al2O3/Si. The LiCO3− trode/electrolyte interfacial impedance.[25] Similar observation signal intensity was much lower on the Al2O3/Si than on Si, was reported by Wang et al.[26] who proposed a solid superacid indicating that little Li2CO3 was observed from the SEI on the model and stated that the co-existence of AlF3 and Al2O3 is benAl2O3/Si electrode. eficial for improving the ionic conductivity and the ion transSeveral reactions have been proposed for the formation of ference in the SEI layer. That might also help explain why the LiF from LiPF6 salt, including salt decomposition:[16,17]
Adv. Mater. 2011, 23, 3911–3915
© 2011 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
wileyonlinelibrary.com
3913
www.advmat.de
capacity of Al2O3/Si is lower under constant current comparing with the level of the uncoated one in our case, and eventually it increases close to the uncoated one after a few cycles. One possible mechanism of the AlF3/Al2O3 formation is the reaction of PF5 resulting from the decomposition of LiPF6 (as shown in Equation 2) with trace water leading to HF which further reacts with Al2O3 and forms AlF3·3H2O.[27,28] In summary, a simple thin film electrode system is designed to understand the influence of ultrathin oxide coating on the electrochemical behavior of Si electrodes for lithium ion batteries. The multiple functions provided by the ultrathin Al2O3 coating on the electrode surface can be summarized as follows: 1), the structure of the Al2O3 changes to Al2O3/AlF3, providing much better lithium ion conductivity. The presence of LiAlO2 on the top surface helps lithium insertion by reducing the associated energy barrier. 2), the Al2O3 coating can serve as an artificial SEI layer, which is chemically stable but highly Li+ ion conductive. This coating suppresses the chemical reaction between the active material and the electrolyte. 3), The Al2O3 surface coating can mitigate mechanical degradation and improve cycling stability.
9.0×104 (a)
4.4 3.4 2.4 1.9 1.4 0.9 0.4 0 Surface
6.0×104 Intensity / a.u.
COMMUNICATION
www.MaterialsViews.com
3.0×104
0.0 85
80
75
70
Binding Energy / eV
Experimental Section Silicon thin films (75 nm thick) were deposited on a copper current collector by RF magnetron sputtering system. Trimethylaluminum (TMA, Sigma-Aldrich, USA) and High Performance Liquid Chromatography (HPLC) graded water (Sigma-Aldrich, USA) were used as precursors in the atomic layer deposition system (Cambridge Nanotech) to deposit the alumina coating on the Si electrode. Details of the electrode synthesis and ALD procedure have been described in our previous work.[12] The Si and Al2O3-coated Si (Al2O3/Si) electrodes were used as the working electrodes with pure lithium metal foil as both the reference and the counter electrode in CR-2032 coin cells. The separator used was a microporous membrane (Celgard, USA) and the electrolyte was 1 M LiPF6 in ethylene carbonate and diethyl carbonate (1:1 volume ratio, Novolyte, USA). The Biological VMP3 potentiostat was used to test the electrochemical performance with constant current – constant voltage (CC–CV) at a cut-off voltage of 1 mV ∼ 2.5 V versus Li/Li+. A current of 10 μA was applied before the potential reached 1 mV, after which, the voltage was held constant till the current converged to 0.1 uA. The constant voltage hold was also added after the completion of constant current charging at 2.5 V. Time-of-flight secondary ion mass spectrometry (TOF SIMS) analyses were conducted on a PHI TRIFT V nanoTOF spectrometer (Physical Electronics, Chanhassen, MN). Before the measurement, the samples were rinsed with dimethyl carbonate (DMC) solvent in order to get rid of the extra electrolyte and salts. Then, samples were kept in sealed vessels filled with Ar during transfer from the Ar filled glove box to the SIMS. The analysis chamber of the instrument was maintained at a pressure of less than 5 × 10−7 Pa during analysis. A 30 kV Au+ ion source is used for both sputtering and analysis. The analysis area was 50 μm × 50 μm, which was in the center of a sputter area of 200 μm × 200 μm. The sputtering rate was maintained at ∼2 nm/min (calibrated with a 100 nm standard SiO2 wafer. A PHI Quantera XPS Scanning Microprobe (Physical Electronics, Chanhassen, MN) with a monochromated Al Ka source (1486.6 eV) was used for the XPS analysis. The base pressure of the system was
